Head of Technology · Themis AI
Building AI that makes safe
decisions under uncertainty.
I’m Stewart Jamieson — an AI researcher and roboticist working on risk-aware, uncertainty-aware machine learning. I lead technology at Themis AI, making AI systems safer and more reliable by teaching them to recognize the limits of their own knowledge — and how to overcome them.
- MIT Ph.D.
- Best Paper · ICRA 2020
- 10+ papers · AI & robotics
- Reviewer · NeurIPS / ICML / ICLR
For a decade I’ve worked on one question from different angles — how do we build AI systems that understand the limits of their own knowledge? From uncertainty quantification for self-driving vision, to risk-aware online learning and Bayesian inference for autonomous robots, to commercializing that research as reliable, de-risked AI at Themis AI.
Selected research
All publications →Finding the optimal exploration-exploitation trade-off online through Bayesian risk estimation and minimization
Artificial Intelligence (Vol. 330) · 2024
DOIDeepSeeColor: Realtime Adaptive Color Correction for Autonomous Underwater Vehicles via Deep Learning Methods
2023 IEEE International Conference on Robotics and Automation (ICRA) · 2023
arXivDOIVideoMulti-Robot Distributed Semantic Mapping in Unfamiliar Environments through Online Matching of Learned Representations
2021 IEEE International Conference on Robotics and Automation (ICRA) · 2021
arXivDOIVideo- ★ Best Paper Award in Service Robotics
Active Reward Learning for Co-Robotic Vision Based Exploration in Bandwidth Limited Environments
2020 IEEE International Conference on Robotics and Automation (ICRA) · 2020
arXivDOIVideo
Recent writing
All writing →- Technical Deep-DivesActing on Low Confidence: Hedge, Abstain, or EscalateWhen a system is unsure, the cost of being wrong — not the confidence score — should decide whether it hedges, abstains, or calls a human.
- AI Strategy & SystemsBuilding Reliable Systems from Uncertainty-Aware ComponentsReliable AI systems come from how you handle uncertainty across components — not from making any single component better. A short field guide.
Recent talks
All talks →- June 2025Capsa: Making AI TrustworthyT-Challenge Competition (T-Mobile & Deutsche Telekom) · Bonn, Germany
- April 2025Themis AI — Startup PitchCSAIL Alliances Annual Meeting · Cambridge, MA
- August 2024Capsa: Making AI TrustworthyAi4 Conference · Las Vegas, NV